Prior to joining a communications network, a wireless device may disadvantageously spend a high percentage of time listening for packets or beacon signals from other devices. Typically, this process involves processing the received signal through a correlation detector that is matched to a preamble or other known start-of-packet delimiter. For example, in accordance with the IEEE 802.15.4 WPAN standard, each packet begins with a preamble comprised of 8 repetitions of a 32-chip pseudo-random sequence. The receiver correlates for this sequence, and then decides whether a valid packet is present responsive to observing multiple correlation peaks. For optimum performance, correlation is performed on multiple frequency offsets in parallel, and the overall power consumption of the correlator represents a significant fraction (10%) of the total receive power. When the preamble is received, the main lobe of the correlation peak is 2-3 samples wide, and it is very unlikely that the strongest correlation peak will be preceded by a very weak correlation value.
Thus, there exists a need for a novel and efficient methodology that advantageously saves power by using a present correlation value to estimate the likelihood that the next correlation value will be a peak. If the likelihood is determined to be low, then the next correlation is skipped, thereby saving power.